AI Agent Handoff and Activities

Agent Activities gives your team one focused place to supervise AI-managed conversations, so you can see what your agents are doing, step in when needed, and hand a conversation back to a human without losing context.

In your inbox sidebar, look for the AGENT ACTIVITIES group near the top, above your regular inboxes. It has three views:

Needs Action: conversations your AI agent has flagged for a human. Agent Working: conversations an agent is actively handling right now. Agent Handled: conversations an agent has already resolved.

Each view is a normal inbox table, so you get the same tools you're used to: search, filters, grouping, and columns for Contact, Topic, Related Cards, Related Locations, and Linked Tasks. Clicking between the three views keeps the conversation list in context, so you can move between "what needs me right now" and "what's already handled" without losing your place.

This is your live view into conversations an agent is currently on. Conversations here show a LIVE badge, and opening one shows a banner at the top of the conversation: "AI is handling this conversation," with a Pause AI button.

Click Pause AI if you want to step in yourself, whether that's because you're keeping an eye on a specific conversation or a customer situation needs a human touch before the agent would normally hand it off. Every message the agent sends is labeled "Handled by [Agent Name]," so it's always clear which messages came from the AI and which came from a teammate.

Once an agent resolves a conversation on its own, without needing a human, it moves here. This is your record of what the AI handled end to end: useful for spot-checking quality, or for understanding how much volume your agents are taking off your team's plate.

This is where an agent hands a conversation back to your team. A conversation moves to Needs Action when:

Because the conversation keeps its full history and any related card, location, or task links, whoever picks it up from Needs Action has everything the agent already gathered, no starting over. Agent will add notes depending on why it triggered a needs action reply. Agent notes are displayed like a comment in the conversation.